Any news on Greek HD camera?

[TrueIndigo]

Does anyone know anything about an Athens-based company which planned to market a series of affordable uncompressed HD cameras using 2/3” to 1.2” sensors? John Papadopoulos posted some interesting information on the DVInfo forum, but since the thread disappeared there's nothing. John said they would be putting up a website -- I've periodically been Googling “ILB16 camera” (the provisional name), but turned up nothing. Anyone have any news on this; it sounded very promising.

[Nicky]

I was also following that cam closely,... please let us know if you find out anything.

[TrueIndigo]

Thanks for replying, Nicky; I was beginning to think I was the only one!


With all the different HD solutions coming out (serving different market sectors from 4k to point-and-shoot), I found the careful choices of this particular camera maker came closer to my own thinking than anything else I'd read. Favouring large pixels instead of chasing the resolution numbers in recognition of the importance of sensitivity. To have grow-able lens options to allow you to start off with C-mount or any manual SLR lens you already have, and maybe moving on to PL mount later. Using affordable RAID laptop drives (inside refillable cartridges) instead of expensive memory cards. And you weren't left to fend for yourself in finding software that could handle the footage. All of this was very attractive and a real candidate for the low budget HD revolution.


I'm hoping maybe John or Otis might see this and chime in with some project news or a link to the company website.

[manymoosh]

Any news?

[TrueIndigo]

Sorry Manymoosh, nothing new. It could be the team behind this camera have now decided to bring it to a finshed state before revealing it to the filmmaking community. However, the projected timeline mentioned on DVInfo has probably lapsed, so I don't know what that means for this camera (I hope it hasn't faded away). I predicted that DVXUser would be one of the forums the team might be aware of, but unfortunately there has been no reply to this thread yet (nor a pm to me personally) from anyone who really knows...

[Otis Grapsas]

Hello from Greece,

I started this project. It's not dead but it's definitely sleeping. Due to the urgency of other projects, it's not getting the time it deserves.

[TrueIndigo]

Otis,
thanks very much for replying. I think there are a number of “silent fans” waiting to see what happens with the camera that was proposed on DVInfo. We look forward to any news, website link and an opportunity to engage in a customer-dialogue about the development of your camera if that is welcome. Is it possible at this time to give some estimate of when the camera(s) are likely to be marketed?

[Otis Grapsas]

Otis,
thanks very much for replying. I think there are a number of “silent fans” waiting to see what happens with the camera that was proposed on DVInfo. We look forward to any news, website link and an opportunity to engage in a customer-dialogue about the development of your camera if that is welcome. Is it possible at this time to give some estimate of when the camera(s) are likely to be marketed?

The original plan was to be ready for beta and preorders by july and take it to market by september, but at this time I can't give an exact estimate. It's about 2 months behind schedule.

I guess we can discuss any questions or suggestions here. The whole idea is to have an open product, so there are not many things than need to change really. It will not be locked to one recording format or sensor or storage type. Remember that the solution is based on an embedded windows machine just like many consumer and professional devices, but it finally uses less energy and is smaller than equivalent closed hardware camcorders/decks. The whole idea was to take advantage of the pc component market. This can keep the product scalable, down to consumer camcorder prices. This market will always be there to push the product into new capabilities and fast integration of any new technology.

I'm going against the trend in this, I know. The logical thing in this market has always been to get a part of the market or even monopolise it with your closed product, format, media, accessory, encoded interfaces, lenses, etc. You get the idea, get battery cells or hard disk drives, put them in a box, overprice them by 2x or 4x and try to enforce their use to your clients by checking battery type, using a custom plug, etc. Or limit capabilities for faster depreciation and forced upgrades. The latest ultimate camera is usually considered old news within a year. I want to do the opposite. Open this market, and make it more accessible by building a product than can grow with the customer needs in an affordable way. Limit cost of production by focusing on software and an enclusure built around industrial components of standard measurements. These two elements will integrate the solution and keep it open and affordable.

[Nicky]

Nice, Im soooo glad you guys are still around. And thanks TrueIndigo for keeping us informed:thumbsup:

Please feel free to post news and updates!

[manymoosh]

Keep us posted, and keep up the good work.

True Indigo, thanks for the update.

All the best

[Ooze3d]

TrueIndigo, thanks a lot for the heads up!

Another one hoping to see these cameras on the market soon!!

[TrueIndigo]

Adopting an OEM method is a very sensible approach to digital camera making; maybe too sensible for some manufacturers!


Otis, the last info I read about the sensors to be used indicated a high quality Kodak CCD in 2/3", 1" and 1.2" sizes. Is this still the case, or has the specification changed? Can you tell us what the maximum frame sizes would be for the different sensors if shooting at 24/25 fps?


Many thanks.

[Otis Grapsas]

The system is designed to be open. The camera will be upgradable to larger/better heads when they appear on the market. We support sensors by Sony, Kodak and two very capable CMOS sensors. The camera does not care really, unless the pixel rate is too high for processing and storage. The good thing is both these are evolving as we speak, so the camera can keep growing.

Every sensor model and sensor technloogy has its advantages and the decision must be discussed in combination with budget, low light and DOF requirements (DOF cuts both ways), lens availability and differences in technology such as ccd smear and cmos rolling shutter. We will help the user choose when the time comes by presenting the hard facts.

Frame sizes do not have significant changes. 720p to a little higher than 2k.

[Aroon_Narayanan]

I know it's probably too early to tell, but what general "ballpark", if you will, are you aiming for in terms of price?

[Otis Grapsas]

Pricing will be in camcorder range, not digital cinema range. 5,000 to 10,000 euro for most models for a full kit (power, storage, monitoring etc).

[Otis Grapsas]

Sony chose CCD and 1920x1080 output on their high end F35 camera.

Sony is very strong in both CCD and CMOS technologies. They manufacture very large CCD and CMOS sensors and have years of experience in both. Their CMOS technloogy variants are very good and they have introduced new technologies that make the best of CMOS. Sony EX1 is excellent for a 1/2" sensor. Sony now only use CMOS in their most affordable cameras. Their 1/3" CCD models are on the way out, and only their most expensive eng and digital cinema cameras use CCD.

I'm happy to see the Sony engineers and customers appreciate CCD motion/global shutter quality and saturation capabilities for their high end Cinealta designs such as the f35. It's aso encouraging that they chose 1920x1080 output and didn't opt for CMOS 6K or 8K or whatever the megapixel hungry crowd demands.

I have to admit the Sony approach makes us feel good about our CCD based digital cinema efforts that start at 1360x820 (eurocine aspect) and end at 2k.

[TrueIndigo]

Otis, I find this very interesting. What I read on DVInfo turned upside down what I thought I knew about cameras and sensors (or should I say, what marketing had directed me to think was best). From the point of view of sensitivity and being able to shoot in natural light, I like the idea of large sensor pixels, though this tends to sacrifice resolution (compared to a same size sensor with smaller pixels). But as John wrote: if you start with an uncompressed source, resolution becomes irrelevant! And larger pixels takes the pressure off lens quality, so it's possible to get great results with lenses that normal people can afford to own. Can you tell me, will the output be 8-bit depth or higher? John mentioned that colour balanced 8-bit was comparable to 10 or 12-bit.

[Nicky]

Yup, very interesting reading their at dvinfo. Why cant we simply have large sensors with few pixel count for good low-light/ dynamic range/ etc. 2K is more than enough but just make sure its raw PLEASE!

[Otis Grapsas]

TrueIndigo,

12 bit is required for alias free 8bit output because resolution is lost in gamma and white balance processing.

Lets assume you are recording using warm incadescent lighting. All colors are on the same AD because we are using bayer like most digital cinema cameras. Your blue will be very low compared to red because you don't want red to clip. You need to boost blue a lot to balance. If you start with 8bit, you will be left with 5 or 6bits of blue. So you need at least 10 bits to be safe from aliasing on your 8bit ouput. Then you have gamma to apply, which wastes another 2 bits in practice due to nonlinear coding. So you do need 12bit when you are going to do these things. Every single camera these days usess 12bits, even when it's ouputing grossly compressed mpeg4 8bit or low quality jpegs.

So, when you start processing white balance and gamma you absolutely need the 12bits to have alias free output. We have an option to do it in camera, so you do not need more than 8bits to be recorded. If you choose to do it in the codec in your editing application (like other "raw" cameras that are actually very compressed), you need to have the full 12bits. It's as simple as that. It does not make much sense to keep 12bit once you have already done white balance and gamma, so nobody does that really.

We do have an option to save up to 16bits linear instead of 8 gamma corrected bits with a hit on storage times and requirements. We don't have cameras that have 16bit ADs, but we do have 14 and 12bit cameras so that would be the effective resolution. It would serve us well in marketing to have higher precision compared to the competition, but it does not offer an advantage really. We recommend 8bit gamma corrected and white balanced output. It is practical and saves storage space. It does not have an impact on quality.

Nicky,

By raw I assume you mean bayer instead of color? It makes sense to save space and that's what most cameras do. If you mean fully uncompressed, we do that also.

The real problem is lenses. There is not much use in having a 1.33" sensor if you are limited to f2.8 optics (nikon still lenses or affordable zooms such as the ones bundled with digital cinema cameras) and you can't afford the f1.3 lenses the 50million $ budget films are shot with. How many $600 DSLRs users have a full series of L series lenses? An entry level Nikon DSLR will usually be used with the slow 18-70 bundled zoom. Which has 2/3" f1.4 DOF capabilities as I explained. If the user is interested in low ight and shallow DOF, he/she needs to spend quite a lot on glass. It's the same with these cinema cameras but the lenses are not as cheap as L series lenses! The entry level bundled zoom can be 8,000 euros. A camera body is one thing. A usable system is something that costs a lot more. A very large sensor is not something the typical DVX user should jump to without thinking, even if the camera body looks affordable.

This is the reason I think very large sensors do not make much sense. Most people getting these cameras want a 10,000 euro complete system but think they can get something better by getting their dream camera. I can provide larger sensors but the slow affordable lenses will limit its application. Once we get passed 2/3" problems get in the way. Cost of lenses is a huge factor.

1.33"(=4/3") f2.8 is identical to 2/3" f1.4 in DOF and sensitivity if same number of pixels and same technology are used on both sensors. 4x larger pixel area but also 1/4" the light due to f2.8 vs f1.4. The same applies to 2/3" f2.8 vs 1/3" f1.4.

A 2/3" with f1.4 primes/zooms has better shallow DOF capaibilities than DSLRs like nikon D70 or Canon 40D with their bundled f3.5 to f4.5 zooms. The same 2/3" also comes close to the DSLRs with f2.8 lenses. 2/3" f1.4 is equivalent to super35mm f3.8 in DOF but A LOT BETTER in light gathering, and a lot more sensititive if we take into account that it has less pixels than the DSLR.

Most people I have talked to want an upgrade from DV, HDV and 1/3" / 1/2" cameras. Not jump from a 6,000 euro DVX setup to a 60,000 euro setup. There is nothing affordable in a 40,000 or 60,000 euro solution. Most people I talk to want to shoot a feature with that kind of money, not buy a camera they cannot afford to own. Renting makes more sense.

[Nicky]

Im obviously not up to date with your developments cos I remember talk of full inch chips? Doesnt matter, things change anyway and 2/3" is still cool like you said with fast affordable glass.

At the moment Im shooting with a Vistavision sized 35adapter (Letus Extreme)combined with full set of Nikon AIS primes f1,4 - f2,0 (around $1200) and its great! Alot of RedOne people are shooting with those lenses - doesnt seem like a big deal.

All I can say is I LOVE shooting wide open with insane DOF. I cant say Im completely understanding the technicals but why cant we have eg. Nikon mount - full sized chip - raw recording at 2K?

[Otis Grapsas]

We have 1" and 1.2" available (1080p and 2k) of course. But most of the interest we are getting is around 2/3". Working professionals and film schools with a stock of 16mm lenses, etc. Not everybody loves the shallow DOF. Getting usable DOF while maintaining sensitivity has its value also. 2/3" f1.4 is close to the limit of usability for most professionals I have talked to. They also use f4 or higher on super35 in most shots, which is almost identical, and only resort to larger apertures when sensitivity is not enough and heavier lighting is not practical. Having only the eyes of a moving subject in focus while following focus is not an ideal situation. Having more in focus does not usually spoil a shot, having a little less forces you to stop down and use more light, which comes at a cost.

With the high sensitivity of the 2/3" you can use DOF adapters and still have higher senitivity than the full frame sensors.

Shallow DOF is a strange situation. The seasoned professionals always wanted deeper DOF and love 2/3" (F900,Varicam etc) for this quality, but those coming from consumer and prosumer 1/3" f1.6 or f1.8 limited camcorders want the shallowest possible in all shots. A teacher at a local film school actually calls it a syndrome:) f1.8 1/6" to f1.4 2/3" is quite a shock, many students want the shallowest dof possible at every singe shot and compromise their result.

To see how 2/3" f1.4 would limit you DOF, check a crop factor DSLR (or DOF adapter) at f4 and remember that in return for the look of f1.4 DOF you are getting an advantage of 8x (3 stops) in light gathering for a given DOF requirement. In other words, if you need 50cm of DOF for your shot at a given DOF/perspective requirement and have a super35 sensor, you are in for changing those 1kwatt lights to 8kwatt and increasing your cost. Things like that matter in low budget shooting.

[Nicky]

I guess thats why everyones shooting 16mm film :) I admit coming from 1/3" cameras we tend to overkill the DOF at first but at least we have the creative option.

I didnt know that smaller sensors are more sensitive to light? I thought it was the other way around?

Quote "Shallow DOF is a strange situation. The seasoned professionals always wanted deeper DOF and love 2/3" (F900,Varicam etc) for this quality, but those coming from consumer and prosumer 1/3" f1.6 or f1.8 limited camcorders want the shallowest possible in all shots. A teacher at a local film school actually calls it a syndrome f1.8 1/6" to f1.4 2/3" is quite a shock, many students want the shallowest dof possible at every singe shot and compromise their result."

At the end of the day, we want to tell a story with visuals as best as possible. Sometimes we want shallow DOF, and sometimes deep DOF - thats for us creatives to decide :)

Im glad to hear youll have larger sensor options.

[Otis Grapsas]

Size of sensor is not very important in sensitivity. The true parameters are quantum efficiency of sensor, pixel size (size vs resolution), large lens aperture availabillity for the sensor format, inherent sensor noise and low light color saturation. The 2/3"sensors have quite large pixels compared to the super35mm sensors because the pixel count is much smaller than the multimegapixel super35mm equivalents. The 2/3" in particular also has the highest quantum efficiency and the highest red/nearIR sensitivity of any chip I have ever seen. This really helps sensitivity and low light shooting which tends to use warm lighting. We are also talking about CCD which means the low light color quality is superior due to higher color saturation out of the chip and less color correction required. The inherent noise of CCD is also very small compared to CMOS chips which helps signal per noise. This is not very important if we have enough light to saturate the sensor (daylight) but becomes extremely important in a light limited situation. Low light quality is extremey important in this design.

[Nicky]

Thanks for explaining, Im always learning new things round here, cant wait to see your product.

[TrueIndigo]

Just thinking ahead about a possible post workflow, can you give some idea of the storage requirements the cameras will need? For example, how big is say a one minute avi file produced by the different sensor sizes? I think it was mentioned before the avi camera file can be used directly in most NLE -- does not require extra proxy files? And there is a performance/resolution playing choice with the codec to allow work even on an older editing machine like a P4 (I'm still using a P4 for editing myself!). Is the sound saved within the avi or recorded separately (presumably as a wav)?
Thanks.

[Otis Grapsas]

I have tested a few different things. Sync audio is recorded inside the avi file. The first way of recording is to get uncompressed bayer and requires 26 to 52mbyte/sec. This can be decoded to color in real time using the codec. Depending on the debayer quality setting this requires a lof of power for full res editing. 1/2 res and medium quality setting is fast enough but still requires some storage thgouhput for 1080p (raid0 for practical editing). 768p is fine with any single modern disk though and uses slightly more space than dvxproHD 1080p and HDCAM.

A friend proposed simple proxies for editing. This is easy to implement. The full res material is recorded on one storage medium and a proxy is stored internally to the camera or through gigE to external media/computers. The proxy uses simpler debayers, can be HD motionjpeg which is very easy on editing resources and quite small and has quality and sizes similar to the compressed dct codecs. It can be other resolutions of course to make editing in older machines even easier.

The original full res uncompressed material can be rendered to the format of choice and produce the final required output at high quality to replace the proxies.

I have also tested wavelet compression on both bayer format (just like other digital cinema cameras) and rgb format with real time debayering. Wavelet requires some serious editing cpu power. I don't believe in this method because fully uncompressed has quality advantages and uses less cpu power.

I believe the proxy method will prevail but some users will want to convert the bayer material to uncompressed video to edit that, and others will want to realtime the bayer online at full res using the codec. I will try to cover all possible workflows.

[TrueIndigo]

Otis, I'm currently using a couple of Win XP machines for my HDV-shot workflow. But I don't fancy Vista much (and maybe not even after a service pack), so I'm looking to eventually migrate to Linux. I installed a distro on an old machine a while back but didn't have enough free time to use it much. Now I'll start finding out what Linux software alternatives there are over the next year or so to make my future platform choice. Can your avi codec be used in a non-Windows NLE?

[Otis Grapsas]

If there is enough demand I will port it and test a coouple of popular editors. There is always the option to convert to a compatible format though.

[TrueIndigo]

Otis, will you have a website up soon? I'm looking forward to seeing more sample frames from the camera. John previously posted stills taken in unfavourable lighting situations and deliberately using a cheap SD lens; that was pretty brave stuff. Made me wonder what it would be like using a good lens like the affordable Fujinon or a classic cine lens (like Angenieux or Schneider). Would also be good to see pictures of what the camera looks like; are you close to having a production camera body fabricated?

[Otis Grapsas]

We are writing material for the web site. I'm in the middle of evaluating a new sensor so that could take a while.

I shot most of the samples Papadopoulos posted under these conditions, because I wanted to show the bare minimum someone could expect from our solution. These were using a very early debayer algorithm, so they were very soft. Most of them had no color correction, since that wasn't implemented at that time. CCDs color is very good without any processing. The resolution using the new algorithms is about 20% more. The lenses were cheap cctv lenses to show that they can produce usable results free from aberations, distortion etc. Modern 16mm cine lenses, primes and the latest megapixel lenses are much better of course but even $250 SD zooms produce usable results. Good SLR optics are also excellent because the best part of the lens frame is used and the pixel sizes are actually larger than the pixel sizes of modern SLRs. Very old cine lenses (the $200 second hand zoom kind) do not have much resolution really because the designs were primitive and many modern technologies were not present when these were designed. But if they look ok with film, they will look fine with CCD. These lenses have some interesting looks. Looks you cannot get with ultra corrected modern designs even if you use a stack of filters and post processing. Not everybody is after the sharpest results so that's a good option to have. Some people are after the Quay brothers romantic look or the Guy Maddin aesthetic or anything of that kind.

Generally, I'm not a fan or extreme resolution. The two known studies on projected 35mm reference material produced resolution results in the 700-750 lines per picture height range so the resolution of our lowest resolution model will not look soft when projected. From an engineering and practicality point of view, that's all I need to know really. I'm more concerned about color quality and pixel quality, so I'm maximising that by using CCD and fully uncompressed recording. I also care about flexibility in fast or even unsteady/creative camera work so I chose global exposure CCD and uncompressed recording for this reason. In other words, I'm trying to solve the limitations I encountered when using affordable cameras.

I'm building another prototype. This time it is much smaller, uses industrial connectors etc. It will still not be the final production design though. Just a concept to feed the final design.

This was shot using a $50 25mm manual cctv lens. It's a 720p 16:9 crop from the 2/3" sensor. Large pixels really solve a lot of problems with optics selection. This early morning sample uses the latest debayer algorithm and no color correction. It's as sharp as many good 16mm cine lenses but not as usable due to small size (similar to the bolex primes). A megapixel lens is twice as sharp. But the user will have to decide if ultra cheap lenses are good enough or cine optics are needed. I think even $50 cctv and $150 machine vision primes or $250 zoom SD lenses and old/affordable cine optics can produce usable results on a large pixel HD camera.

http://img175.imageshack.us/img175/5760/a0001xj9.jpg

[lawriejaffa]

This sounds like a very nice project Otis - I'll check the earlier posts for news about this product.

It seems you are developing this with a sense of caution and humility (that makes a nice change from RED sensationalism hehe).

The pic looks nice - you've been asked this a hundred times im sure but when do you hope to aim for a release of this product? It would probably be a special order item!?

Anyway keep up the good work!

[Otis Grapsas]

Thanks Lawrie.

The system still needs a few months of development.

I haven't worked out the logistics of production yet. We will do our best.

[TrueIndigo]

Hey, Guy Maddin is one of my favourite directors! Regarding older cine C-mount lenses, I have an f1.8 10mm Shneider, a f1.3 15mm Angenieux and an f1.9 26mm Bausch & Lomb which I am looking forward to using again. Someone once said these older lenses have “lots of character”, it's something difficult to describe (like analogue amps versus digital in the music industry). I'm not against what the latest technology has to offer and a new lens from the reasonably priced Fujinon range would probably be the next type I would like to try (is it a Fujinon you will be supplying with the camera kit? What focal length?).

Thanks for the latest information.

[Otis Grapsas]

Megapixel primes are a hot market at this time. Many new designs are out using modern technology (well corrected, very sharp, free from distortion at wide angle with lots of aspherical elements, floating optics, etc). The 2/3" sensor size gets the most options. I will bundle a very sharp 12mm which is a normal at 2/3". I would love to see someone get through a full project using just the 12mm:) He/she would learn a lot in the process.

A prime lens kit for the 2/3" would give the equivalent of the following lenses in 35mm full frame still photography (while producing a sharp image down to 5um pixels):

20mm f2.8 0.5% distortion
32mm f1.4 0.6% distortion
50mm f1.4 0.07% distortion
65mm f1.4 0.05% distortion
100mm f1.4 0.07% distortion
140mm f1.4 0.05% distortion
200mm f1.4 0.03% distortion
300mm f1.4 0.05% distortion

A good set would be 20,32,50,100,200 which will cost about 1000 euro.

It's interesting how affordable large aperture availability helps DOF and light gathering.

75mm f1.4 on 2/3" produces a 200mm f3.7 image equivalent in super35mm (identical angle, geometry and DOF) .

A super35mm sensor will probably have to settle for 200mm f2.8 because larger apertures cost a lot in cine lens and are very rare in still lens. As an example, the RED lenses are f2.8.

So the 2/3" will have just a little deeper DOF (f3.7 vs f2.8) but it will have a light gathering advantage of 4x (2 stops, f1.4 vs f2.8) compared to the super35mm.

This also applies in most telephoto cases. f2.8 is a limit for slr optics above a point, and the cine lenses (even some expensive bundled ones) are also limited to f2.8 once we get into telephoto range. But the 2/3" sensor can use large aperture machine vision and 16mm cine optics and also large aperture normal/portrait lenses from SLRs which become telephoto (50mm f1.4 85mm etc). The real question is: How much of the telephoto super35mm shallow DOF do you actually get with the lenses you can afford? 2/3" will only give a slight loss of shallow DOF once you put the lens into the equation, and the light gathering advantage is impressive. Then we have the 2/3" working to serious advantages when a given DOF is required and light must not be lost such as in any case where the camera moves a lot in low light and focus pulling becomes a problem.

[Otis Grapsas]

http://www.archive.org/download/TwoBottlesInAndOutOfFocus/bottlesinoutoffocus.wmv

An example of using a cheap large aperture SD industrial lens. Focus pulling by hand. This was shot months ago and is obvioulsy very compressed. The color is not balanced and the shadows were clipped by mistake. This was shot with available light, a single 60watt incadescent on a ceiling fixture. Generally, it's not represenatative of camera quality.

This is the lens. It costs about 80 euro:

http://www.bhphotovideo.com/images/largeimages/546271.jpg

This 25mm f1.4 on 2/3" is equivalent to an 67mm f3.7 lens on super35 so it still has some shallow DOF. The good thing is that the DOF is deep enough to just cover a head (nose to ears at least) but still gathers light as an f1.4.

The distance from bottle to bottle is about 40cm. The camera is about 60cm from the bottle, and the bottle in the back is about 50cm from the wall.

[semila58]

Otis, you mentioned that the sensor could be swapped. Is anybody making 4:3 sensors that could be used in it? That would make it the only 4:3 HD camera besides the the Arri D-20/21 and DVX Andromeda mod. That would be really cool.

And hey! All my test footage has whiskey in it to!:beer:

[Otis Grapsas]

Otis, you mentioned that the sensor could be swapped. Is anybody making 4:3 sensors that could be used in it? That would make it the only 4:3 HD camera besides the the Arri D-20/21 and DVX Andromeda mod. That would be really cool. :beer:

We are not swapping just the sensor, that would require a global support circuit which is almost impossible to manufacture and hard to optimise for a specific sensor not to mention that it would eventually get outdated. We are accommodating OEM circuitry, modules using widely used interfaces and coming with many different sensors. That means it's almost plug and play to upgrade. Only a software update is needed. We can probably accommondate any future sensor technology with minimal effort this way without passing a tremendous new development cost to the end user. The rest of the camera system stays as is and only a new software pipe for pixel input is linked, and software tuning for the specific sensor is performed.

Most sensors are 4:3 and some very good ones are 1:1. 16:9 is actually rare and only used in the video market. 16:9 and wider than 4:3 can always be produced by a crop.

This particular 4:3 sensor is 1392x1040 which makes it same resolution with 1920x1080 on 4:3 crop (which is 1440x1080).

I have thought about ways to make use of the extra vertical area that comes with these sensors.

The first use is to have extra area vertically for live monitoring of boom mics and anything else that is close to the frame edges.

The area can be used to change vertical framing in post. Widescreen is sometimes problematic in that department.

The extra area can also be used to stabilise material in post. Material shot from moving cars, dollies etc has a lot of unsteady motion in the vertical axis, so having an extra 30% or so helps maintain vertical resolution.

Anamorphic optics can also be used, but I'm not familiar with their specifics.

Then, a number of people actually want to shoot in 4:3, especially in black and white to get a different framing/composition or simply an old cine or tv look.

And hey! All my test footage has whiskey in it to!:beer:

Now that you mention it, if I knew we had experts in the audience, I would shoot Lagavulin 16s or any brand that actually has some class:)

[semila58]

Okay, thanks for the clarification on the sensor. And as far as the whiskey goes, I had just assumed you were saving the Lagavulin for the big release date. Looking forward to it!

Peace,
S-

[Otis Grapsas]

To anyone interested in this project:

We will not get into preorders until we are very close to final production stage, complete with testing and a working, practical, productive and affordable workflow.

Thanks.

[j]

Keep us posted, there is interest.

[ryan brown]

Keep us posted, there is interest.

:smiley_up:smiley_up what he said.

[Horncastle]

Keep us posted, there is interest.
I second that. And this is a very informative thread too - thanks Otis.
Jason

[Postmaster]

I third that Otis. I´m verry intrested.

And I learned more about sensors in the last 20 minutes, reading your coments,
than in the last 2 years.

Thanks.

Frank

[TrueIndigo]

The recent shock of DSLRs shooting video has set the forums alight. I honestly can't remember such a high level of excitement over what it represents (that maybe small sensor camcorders and 35mm adapters may have been made obsolete almost overnight). However, when I saw hundreds of enthusiasts on Mr. Laforet's blog literally begging Canon to give them something as fundamental as 24 and 25 fps, it made me realize the gulf between a DSLR with a video facility and a purpose-made camcorder. The things people are already unhappy with (rolling shutter, manual controls, image codec, integrated sound) are not an issue with a camera like the ILB16. And where the DSLRs are impressive (low light, shallow DoF) there is every expectation the various ILB16 models will be very competitive on these specifications (neither the Canon 5D2 nor the ILB16 are released yet, so no real world comparison can yet be made). A video DSLR is very portable and attractively priced, though to make them seriously usable you'll probably need a shoulder mount, TFT monitor, separate sound solution, plenty of extra batteries and memory cards. If you don't already own that stuff this can more than double the price, bringing you back into the realm of the ILB16 (whose kit includes all of these DSLR extras). Some people on the forum are doing great work trying to figure out how film makers can work round the limitations of these video DSLRs, and I will follow it with interest. I just think, according to my own current analysis, I'll be following this camera with more interest.

[yzmotoxer807]

Also very interested! :)

And I totally agree with this too:
And I learned more about sensors in the last 20 minutes, reading your coments,
than in the last 2 years.
Frank

[Otis Grapsas]

Thanks.

The purpose of the project is to improve low budget production. Imaging is just a small part of the equation. My experiece with camcorders in this range convinced me about that! Lack of 24p, lens quality, interlacing, pixel shifting, high compression, rolling shutter, low light performance and deep dof were disadvantages and are gradually solved in newer models, but so was cost of quality video monitoring, quality of audio, limited audio feature set (and compression), quality and durability of connectors, transfer and backup times, form factor, cost of media, availability of media, camera upgradability, availability of ultra sensitive b&w sensors, etc.

The dsrs have excellent imaging sections and very good processing and color. The power was there so the manufacturers offered it. They will limit performance by not proving high quality i-frame compression or 24p or good audio quality or control of exposure. I don't think we will ever see CCD on dslrs again because cmos is cheaper, rolling shutter is cheaper, has no problems in still imaging, and has technical advantages over global shutter.

[combatentropy]

hi otis, I'm interested in your project.

if you will, please give me your feedback on a camera design I made recently

http://www.combatentropy.com/a_cam_for_all_seasons

[Otis Grapsas]

It's very good. But gobal shuttering at this price is impossible. If it was CCD with this sensor size it could easily cost 30,000 just for the imaging part and if it was global shutter cmos, it would be no less than 10,000 for imaging. So we are talking about a 100,000 or a 50,000 $ camera. For a 5000 $ camera with global shutter ccd you would have to settle for 1/2" ccd and slightly lower resolution. It can certainly be done at this price using a 2/3" rolling shutter cmos but it would take 15,000 or 20,000 sold units to break even in these cases, especially with a fixed lens.

[Otis Grapsas]

I have tried to keep the cost realistic. Realistic in cinematography means less than 20,000 euro for a full kit that is capable of supporting production of a real feature. Including mics, a basic rig, tripod, audio, lights, lenses and an editing station.

Many low budget cinematographers have taken 1/4" or 1/3" dv to that cost once the required elements are put in the equation, just to discover they are limited by the camera, so that would be revolutionary for the higher quality imaging and audio and the flexibility provided with our camera. 20,000 euro might be a lot for many people, but the kit can be shared by a couple of cinematography students for their projects and owning the equipment gives a lot of power and flexibility.

That's the plan. Something the dv budget can reach to. I can do a large sensor camera instead, an affordable one also, but that will come with cmos, low light performance that requires lighting for film and large crews, a focus puller, expensive lenses, etc. Not realistic in my book. Needing 600watts instead of 100watts is a big deal in low budget shooting and artistically limiting in many cases. I'm building a camera that will reduce cost of shooting by saving time and reducing demand for a larger crew. My purpose is not a price tag that will appear low at first glance but require expensive production. It's to create a camera that is cheaper than dv in actual use due to its low light capabilities while providing the features a cinematographer needs.

I have to add that my studies have nothing to do with programming. I studied civil engineering with a project management specialisation in university. The goal of my studies is to take a project having limited resources (even 2 billion euro projects have limited resources) and put it through production effectively, optimise resource use and maximise product quality. It's exactly what filmmaking is in practice. I have a passion for audio and video but I have also learned to keep production realities in mind. A product that looks good on paper, from a marketing perspective or on a show display is not very important to me.

[karapetkov]

Hey, Αθήνα is closer that Stockholm. :)

I might just make a "digital cinema" trip down south.

Do you guys happen to have an office in Thessaloniki?

Gas isn't cheap these days, ya know. ;)

[Otis Grapsas]

We are at garage stage since we are still developing.

We don't have a garage in Thessaloniki:)

[TrueIndigo]

"Optimise resource use and maximise product quality" -- yes, I'm not convinced that "convergence" will produce the right tool for the job. If you're going to buy a new DSLR anyway, great: now you'll get a free video camera thrown in -- I think that's the original spirit in which the video facility is being offered. But if you are buying say the Canon D5 mk 2 just for the video, the problem is you are also paying for an expensive full frame DSLR as well -- that money would be better spent totally on the video capability of your camera. I believe there will always be a place for a camera which does just one thing really well. There are enough compromises in low budget production as it is without adding to it!

[combatentropy]

If it was CCD with this sensor size it could easily cost 30,000 just for the imaging part and if it was global shutter cmos, it would be no less than 10,000 for imaging.

I don't understand. Why would it cost so much? Is it because manufacturers sell so few of them?

It's only an extra 1.5 square centimeters of silicon. So it couldn't be for the extra material. And the complexity is the same as a 2/3". In fact, I would have guessed that the 2/3" would cost more, because it has the same complexity (1,920 by 1,080 pixels), but everything is made smaller.

[combatentropy]

I'm not convinced that "convergence" will produce the right tool for the job. . . . I believe there will always be a place for a camera which does just one thing really well. There are enough compromises in low budget production . . .

The belief that every job takes its own kind of camera is something I hear from several posters, but that is also something that big camera companies push on us. A highly segmented market gets them more profit.

I have shot about 200 projects over the past 15 years: fictional stories, commercials, promotionals, weddings, concerts, and so on. I feel like I have a sense of the ideal set of features. Yes, there may be a place for a 4K camera if you want to shoot an epic, or a super-high-speed camera for special effects. But the market is way more segmented than it needs to be. There are many who want to shoot the occasional independent film, and make some extra money on the side by shooting a wedding or a concert, and do it all with a camera that they own. For now, the best candidate is probably something like the HMC150.

Otis, you're on the right track. There must be compromise, but a better balance can be struck if we (a) avoid greed and (b) design smartly.

[Otis Grapsas]

I don't understand. Why would it cost so much? Is it because manufacturers sell so few of them?

It's only an extra 1.5 square centimeters of silicon. So it couldn't be for the extra material. And the complexity is the same as a 2/3". In fact, I would have guessed that the 2/3" would cost more, because it has the same complexity (1,920 by 1,080 pixels), but everything is made smaller.

Full frame is 16x greater area compared to 2/3" and flaws can quickly reduce high grade output. Not to mention that the same area can produce dozens of consumer digicam sensors. A good fast 2/3" ccd is already 500 euro without support electronics (very expensive in CCD for video rates). High grade sensors are very expensive in large sizes, a multiple of the price of the consumer grade. The cameras will be harder to design thermally, they need more expensive bodies and electronics, multiple ADs etc.

The global shutter CMOS solutions have more transistors I believe. It's also a very small market because rolling shutter does the job in multiple applications.

[combatentropy]

Full frame is 16x greater area compared to 2/3"

My design has a Super 35 sensor, not a "full frame" 35mm still format sized one.

2/3" sensor = 10.6 mm x 6.9 mm = 73.1 sq. mm. (Kodak's (http://www.kodak.com/global/en/business/ISS/Products/Interline/KAI-02150/overview.jhtml?pq-path=13614))
Super 35 mm sensor = 24.4 mm x 13.7 mm = 334.2 sq. mm. (RED's (http://www.red.com/cameras))

334.2 / 73.1 = 4.5

The difference between Super 35 and still-photo 35 confuses many, especially since Super 35 is the smaller one. A quicker and clearer way I suggest we all refer to either is S35 (http://en.wikipedia.org/wiki/Super_35) (Super 35) or 135 (http://en.wikipedia.org/wiki/135_film) (the official name for "full frame" 35mm).

[Otis Grapsas]

The video capable 5D II adds to the confusion now:)

[manymoosh]

Otis, wonderful thread. Really enjoyed reading it (learned alot).

I hope your up and running soon, keep up the good work!

All the best,

[Otis Grapsas]

I just did a test using the 2/3" 1920x1080 ccd the A-Cam is supposed to use using my prototype. I had done tests using the older 1" 1920x1080 version, so this was a drop in replacement. Another plus for the modular design. Support for the new Kodak took less than an hour. I still prefer the 768p 2/3" over this one, but it's 1080p for those who do not feel safe with lower than 1080p.

The output looks great so we are supporting it.

[hawaj]

... any framegrabs to post Otis?

[Otis Grapsas]

I will refrain from doing that at the moment:) At least untill we see something from the A-Cam final implementation.

[TrueIndigo]

How about this as a slogan: ILB16...controversially sensible cameras.

[Otis Grapsas]

We will market it as a system, not just a camera. It can do many things that are typically handled by specialised equipment. I can't disclose the feature set of the system, but the functionality is way beyond any other camcorder, digital cinema or raw recording camera out there and some of the extra functionality is unthought of for a camera.

Lets face it, a raw recording camera can just record digital data out of a sensor block while providing monitoring. We are after a lot more. The goal is a digital cinematography swiss army knife. The main advantage is the power that is provided by the platform we use, the main production tool of this century, the personal computer.

From the communication with interested parties so far, we will see lots of customisation on the product, because it's a system you can build on and upgrade, with lots of options on imaging, audio, storage, communication with the environment, etc.

As an example, it only took a single day and $100 of hardware to add and implement hand cranking (ramping). An example on one of the many things that are impossible or cost prohibitive on the closed specification camera solutions due to their architecture. On our system, the limit is imagination. We have already tested functionality that is very hard or expensive to implement on other cameras, remote control, 3d imaging, GPS functionality etc. Open hardware is power.

[TrueIndigo]

A few more ideas:

Adaptable. Upgradable. Affordable...it's going to take a bit of getting used to.

We know it's your money, your movie, and we're on your side.

Lots of cameras. Lots of components. You have a choice...so do we.

Independent film. Independent camera. Made for each other.

Talking up the facts, not the figures.